Fan high-temperature anomaly detection method and system based on time sequence feature fusion

Abstract

The invention provides a draught fan high-temperature anomaly detection method and system based on time sequence feature fusion, and the method comprises the steps: obtaining Scada data of a draught fan set, carrying out the downsampling processing and trend quantification processing, replacing a boundary instantaneous value, carrying out the smooth high-frequency noise processing through moving average filtering, and extracting a temperature feature and an operation state feature; physical mechanism rules are associated, and mechanism modeling is carried out; coupling characteristics of the Scada data are obtained through mechanism modeling or calculation, wherein the coupling characteristics comprise a device-cooling temperature difference ratio (TCR), a load-temperature gradient matching degree (LTM) and a physical constraint conflict mark (PCC); and distributing different weights to the coupling features based on the coupling strength to carry out abnormal scoring and multi-level early warning. According to the method and the device, fan performance degradation caused by historical dependence can be avoided, root cause accurate positioning is realized, the operation and maintenance troubleshooting cost is reduced, false alarms caused by sensor noise or transient load fluctuation are filtered, the false alarm rate is reduced, and the accuracy rate is improved.

Description

Technical Field

[0001] This application relates to the technical field of health management of wind turbine generators, and particularly relates to a method and system for detecting high-temperature anomalies of wind turbines based on the fusion of time-series features. Background Art

[0002] Currently, the commonly used methods for detecting abnormal temperatures of wind turbines include the single-threshold alarm method, the LSTM-based prediction method, and the machine learning model classification method. Among them, the single-threshold alarm method detects temperature anomalies by setting a fixed temperature threshold. The LSTM-based prediction method predicts the theoretical temperature through historical data and compares it with the actual temperature value to judge temperature anomalies. The machine learning model classification method uses random forest or support vector machine (SVM) to perform binary classification on instantaneous temperature data.

[0003] However, the above methods have certain limitations. First, the threshold alarm method causes false alarms when the instantaneous temperature suddenly changes, resulting in the system capturing not real temperature anomalies but noise. Second, the LSTM-based prediction method ignores the time-series inertia characteristics of temperature changes, such as the heating rate and cumulative heat, for the judgment of a single time point. In addition, there is no association with equipment industrial control signals concurrent with high temperature, such as variable speed and hydraulic pressure, resulting in the failure-derived characteristics not being fully and effectively utilized. Moreover, after the above methods mark anomalies, manual review of the time-series segments is required, and accurate start and end timestamps cannot be automatically output.

[0004] The above methods rely on statistical algorithms and manipulations. The statistical features have no actual physical meaning, cannot adapt to sudden changes, have a high false alarm rate, and cannot guide maintenance. To address the above problems, this application proposes a method and system for detecting high-temperature anomalies of wind turbines based on the fusion of time-series features. Summary of the Invention

[0005] In view of one or more technical defects in the above-mentioned prior art, this application proposes the following technical solutions.

[0006] Based on the first aspect of this application, a method for detecting high-temperature anomalies of wind turbines based on the fusion of time-series features is proposed, including: S1: Obtain the Scada data of the wind turbine group for downsampling processing, process the boundary instantaneous values in the Scada data through a trend quantization method, and then perform smoothing of high-frequency noise through a moving average filter to extract the temperature features and operating state features of the Scada data; S2: Divide the Scada data into a generator temperature control module, a cooling system module, an electrical load module, and a mechanical state module, and associate the temperature features and operating state features of the Scada data with physical mechanism rules to perform mechanism modeling; S3: Combine mechanism modeling or calculations to obtain the coupling characteristics of the Scada data, where the coupling characteristics include the equipment-cooling temperature difference ratio TCR, the load-temperature gradient matching degree LTM, and the physical constraint conflict flag PCC; S4: Assign different weights to the coupling characteristics based on the coupling strength for anomaly scoring, and perform multi-level early warning based on the anomaly score. When the anomaly score is greater than the preset basic threshold, a yellow early warning is issued. When the anomaly score is greater than the preset basic threshold and the number of physical constraint conflict flags PCC that are 1 exceeds the preset number threshold, a red early warning is issued; The anomaly scoring formula is: , where TCR represents the equipment-cooling temperature difference ratio, LTM represents the load-temperature gradient matching degree, PCC represents the number of physical constraint conflict flags that are 1, Score represents the anomaly score, and Weight represents the different weights assigned to the coupling characteristics.

[0007] Furthermore, the temperature characteristics include the pitch drive temperature, the ambient temperature, the water-cooled outlet valve water temperature, the water-cooled inlet valve water temperature, the nacelle temperature, the generator bearing temperature, and the hub temperature; The operating state characteristics include the pump state and the fan state.

[0008] Furthermore, step S1 also includes filtering the Scada data where the temperature characteristics exceed the physical limit threshold, cleaning the Scada data with logical conflicts, replacing the missing values with the mean of the Scada data, and performing sliding window time series processing on the Scada data.

[0009] By detecting physically infeasible events, removing data that violates the basic laws, and refining an information flow that is physically credible, logically self-consistent, and temporally coherent, it provides high-quality input for subsequent analysis and improves the reliability of subsequent analysis.

[0010] Furthermore, the calculation formula for the equipment-cooling temperature difference ratio TCR is: , where TCR represents the equipment-cooling temperature difference ratio, represents the critical equipment temperature, represents the corresponding cooling medium temperature, represents the minimum value.

[0011] Furthermore, the calculation formula for the load-temperature gradient matching degree LTM is: , where LTM represents the load-temperature gradient matching degree, Indicates the temperature change of the device within the window, indicates the change in load power, indicates the minimum value.

[0012] Furthermore, define the physical constraint conflict flag PCC based on the system physical mechanism rules, including: If the temperature of the front bearing of the generator is higher than the preset front bearing temperature threshold, and the temperature difference between the inner ring temperature and the outer ring temperature of the front bearing or the rear bearing of the generator exceeds the preset bearing temperature difference threshold, then the physical constraint conflict flag PCC is 1; If the water temperature at the outlet valve of the water cooling is higher than the preset water cooling temperature threshold, and the temperature difference between the water temperature at the outlet valve of the water cooling and the water temperature at the inlet valve of the water cooling exceeds the preset water cooling temperature difference threshold, then the efficiency of the cooling system decreases, and the physical constraint conflict flag PCC is 1; If the ambient temperature is higher than the temperature threshold of the high-temperature weather warning line, and the temperature of the engine room is higher than the preset forced ventilation failure threshold, then there is an abnormality in the ambient or engine room temperature, and the physical constraint conflict flag PCC is 1; Otherwise, the physical constraint conflict flag PCC is 0.

[0013] Combining the physical conflict constraint identifier PCC to filter out false warnings caused by sensor noise or short-term load fluctuations can reduce the false alarm rate and guide maintenance.

[0014] Furthermore, the division of the Scada data into a generator temperature control module, a cooling system module, an electrical load module, and a mechanical state module is specifically as follows: The generator temperature control module includes the temperature of the front bearing of the generator, the temperature of the rear bearing of the generator, the inner ring temperature of the front bearing of the generator, and the outer ring temperature of the front bearing of the generator; The cooling system module includes the water temperature at the outlet valve of the water cooling, the water temperature at the inlet valve of the water cooling, the pump state, and the fan state; The electrical load module includes the active power of the converter and the generator speed; The mechanical state module includes the hydraulic system pressure and the bearing vibration amplitude.

[0015] Furthermore, assigning different weights to the coupling features based on the coupling strength includes: If two or more physical mechanism rules are triggered in the current window, assign a high weight to the coupling feature; If one physical mechanism rule is triggered in the current window, assign a medium weight to the coupling feature; If no physical mechanism rule is triggered in the current window, assign a low weight to the coupling feature.

[0016] Associating abnormal types with physical components through dynamic weight allocation can reduce the troubleshooting cost for operation and maintenance personnel.

[0017] Based on the second aspect of the present application, a fan high-temperature anomaly detection system based on time-series feature fusion is proposed, including: Data processing module: Obtain the Scada data of the fan group for downsampling processing, process the boundary instantaneous values in the Scada data through a trend quantization method, and then perform smoothing of high-frequency noise through moving average filtering to extract the temperature features and operating state features of the Scada data; Mechanistic modeling module: Divide the Scada data into a generator temperature control module, a cooling system module, an electrical load module, and a mechanical state module, and associate the temperature features and operating state features of the Scada data with physical mechanism rules for mechanistic modeling; Coupling module: Combine the mechanistic modeling or calculation to obtain the coupling features of the Scada data, and the coupling features include the equipment-cooling temperature difference ratio TCR, the load-temperature gradient matching degree LTM, and the physical constraint conflict flag PCC; Early warning module: Assign different weights to the coupling features based on the coupling strength for anomaly scoring, and perform multi-level early warning based on the anomaly scoring. When the anomaly scoring is greater than a preset basic threshold, a yellow early warning is issued. When the anomaly scoring is greater than the preset basic threshold and the number of physical constraint conflict flags PCC equal to 1 exceeds a preset number threshold, a red early warning is issued; The anomaly scoring formula is: , where TCR represents the equipment-cooling temperature difference ratio, LTM represents the load-temperature gradient matching degree, PCC represents the number of physical constraint conflict flags equal to 1, Score represents the anomaly score, and Weight represents different weights assigned to the coupling features.

[0018] By replacing the statistical model with the physical mechanism rules and the calculated coupling features in the present application, it can directly reflect the physical coupling relationship between the equipment operating state and the cooling system, and improve the accuracy of the pitch drive overheating and cooling pump failure scenarios.

[0019] Based on the third aspect of the present application, a computer program product is further proposed, which has one or more computer programs, and when the computer programs are executed by a computer processor, the method described in any one of the above is implemented.

[0020] The technical effects of this application are as follows: This application conducts high-precision anomaly detection through physical mechanism rules, directly reflects the operating state of the equipment and the physical coupling relationship of the cooling system through coupled features, directly correlates the anomaly type and physical components through dynamic weight allocation, can avoid the performance decay of the fan caused by historical dependence, achieve accurate root cause positioning, reduce the operation and maintenance troubleshooting cost, filter out false alarms caused by sensor noise or short-term load fluctuations, reduce the false alarm rate, and improve the accuracy rate. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] Other features, objects, and advantages of this application will become more apparent by reading the detailed description of the non-limiting embodiments with reference to the following drawings.

[0022] Figure 1 is a flowchart of a method for detecting high-temperature anomalies of a fan based on temporal feature fusion according to an embodiment of this application.

[0023] Figure 2 is a module diagram of a system for detecting high-temperature anomalies of a fan based on temporal feature fusion according to an embodiment of this application.

[0024] Figure 3 is a schematic structural diagram of a computer system of an electronic device suitable for implementing the embodiments of this application. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0025] The following further elaborates on this application in conjunction with the drawings and embodiments. It can be understood that the specific embodiments described herein are only used to explain the related invention and do not limit the invention. Additionally, it should be noted that for ease of description, only parts related to the invention are shown in the drawings.

[0026] It should be noted that, without conflict, the embodiments in this application and the features in the embodiments can be combined with each other. The following will detail this application with reference to the drawings and in conjunction with the embodiments.

[0027] Figure 1 A method for detecting high-temperature anomalies of a fan based on temporal feature fusion is shown, including: S1: Obtain the Scada data of the fan group for downsampling processing, process the boundary instantaneous values in the Scada data through a trend quantization method, and then perform smoothing of high-frequency noise through a moving average filter to extract the temperature feature and operating state feature of the Scada data; S2: Divide the Scada data into a generator temperature control module, a cooling system module, an electrical load module, and a mechanical state module, and correlate the temperature feature and operating state feature of the Scada data with physical mechanism rules for mechanism modeling; S3: Combine mechanism modeling or calculation to obtain the coupling characteristics of the Scada data, where the coupling characteristics include the equipment-cooling temperature difference ratio TCR, the load-temperature gradient matching degree LTM, and the physical constraint conflict flag PCC; S4: Assign different weights to the coupling characteristics based on the coupling strength for anomaly scoring, and perform multi-level early warning based on the anomaly scoring. When the anomaly scoring is greater than the preset basic threshold, a yellow early warning is issued. When the anomaly scoring is greater than the preset basic threshold and the number of physical constraint conflict flags PCC equal to 1 exceeds the preset number threshold, a red early warning is issued; The formula for the anomaly scoring is: , where TCR represents the equipment-cooling temperature difference ratio, LTM represents the load-temperature gradient matching degree, PCC represents the number of physical constraint conflict flags equal to 1, Score represents the anomaly scoring, and Weight represents the different weights assigned to the coupling characteristics.

[0028] It should be noted that the temperature characteristics include the pitch drive temperature, the ambient temperature, the water-cooled outlet valve water temperature, the water-cooled inlet valve water temperature, the nacelle temperature, the generator bearing temperature, and the hub temperature; The operating state characteristics include the pump state and the fan state.

[0029] It should be noted that step S1 further includes filtering the Scada data with the temperature characteristics exceeding the physical limit threshold, cleaning the Scada data with logical conflicts, replacing the missing values with the mean value of the Scada data, and performing sliding window time series processing on the Scada data.

[0030] It should be noted that the formula for calculating the equipment-cooling temperature difference ratio TCR is: , where TCR represents the equipment-cooling temperature difference ratio, represents the key equipment temperature, represents the corresponding cooling medium temperature, represents the minimum value.

[0031] It should be noted that the formula for calculating the load-temperature gradient matching degree LTM is: , where LTM represents the load-temperature gradient matching degree, represents the temperature change of the equipment within the window, represents the change in the load power, represents the minimum value.

[0032] It should be noted that when the load-temperature gradient matching degree LTM > 0.5, the temperature rises too fast, exceeding the load expectation, triggering the physical mechanism rule; when the equipment-cooling temperature difference ratio TCR suddenly increases by more than twice or the heat dissipation efficiency decays during the rise in 3 consecutive windows, the physical mechanism rule is triggered.

[0033] It should be noted that the physical constraint conflict flag PCC is defined based on the system physical mechanism rule, including: If the temperature of the front bearing of the generator is higher than the preset front bearing temperature threshold, and the difference between the inner ring temperature and the outer ring temperature of the front bearing or the rear bearing of the generator exceeds the preset bearing temperature difference threshold, then the physical constraint conflict flag PCC is 1; If the water temperature at the outlet valve of the water cooling is higher than the preset water cooling temperature threshold, and the temperature difference between the water temperature at the outlet valve of the water cooling and the water temperature at the inlet valve of the water cooling exceeds the preset water cooling temperature difference threshold, then the cooling system efficiency decreases, and the physical constraint conflict flag PCC is 1; If the ambient temperature is higher than the temperature threshold of the high-temperature weather warning line, and the cabin temperature is higher than the preset forced ventilation failure threshold, then there is an abnormality in the ambient or cabin temperature, and the physical constraint conflict flag PCC is 1; Otherwise, the physical constraint conflict flag PCC is 0.

[0034] It should be noted that the division of the Scada data into a generator temperature control module, a cooling system module, an electrical load module, and a mechanical state module is specifically as follows: The generator temperature control module includes the temperature of the front bearing of the generator, the temperature of the rear bearing of the generator, the inner ring temperature of the front bearing of the generator, and the outer ring temperature of the front bearing of the generator; The cooling system module includes the water temperature at the outlet valve of the water cooling, the water temperature at the inlet valve of the water cooling, the pump state, and the fan state; The electrical load module includes the active power of the converter and the generator speed; The mechanical state module includes the pressure of the hydraulic system and the bearing vibration amplitude.

[0035] It should be noted that different weights are assigned to the coupling characteristics based on the coupling strength, including: If two or more physical mechanism rules are triggered in the current window, a high weight is assigned to the coupling characteristics; If one physical mechanism rule is triggered in the current window, a medium weight is assigned to the coupling characteristics; If no physical mechanism rule is triggered in the current window, a low weight is assigned to the coupling characteristics.

[0036] It should be noted that the core equipment includes a pitch drive and a generator bearing. The triggering of two physical mechanism rules includes a sudden increase in the equipment-cooling temperature difference ratio TCR and a physical constraint conflict flag PCC greater than or equal to 1. The triggering of one physical mechanism rule means that one of the equipment-cooling temperature difference ratio TCR, the load-temperature gradient matching degree LTM and the physical constraint conflict flag PCC is triggered.

[0037] It should be noted that the values of the high weight, medium weight and low weight decrease in sequence and are configured according to site conditions and actual needs.

[0038] It should be noted that this application is based on the real-time health assessment method of the thermodynamic characteristics of industrial equipment, and innovatively proposes a two-dimensional coupling analysis model of equipment-cooling temperature difference ratio TCR and load-temperature gradient matching. The equipment-cooling temperature difference ratio TCR continuously quantifies the heat dissipation efficiency attenuation characteristics by comparing the temperature gradient distribution of key nodes in the heat exchange system; the load-temperature gradient matching LTM constructs a time-domain criterion for the non-steady-state temperature rise trend based on the dynamic process detection of temperature change dynamics; The two parameters work together to study the evolution law of thermal load of the core components of wind turbine rotating machinery. By extending the time-frequency dimension of the observation continuity window period, the early signs of abnormal thermal state of the equipment are captured without the intervention of prediction algorithms. Compared with the traditional thermal failure warning technology based on predictive models, this application relies on the time series intrinsic analysis of SCADA data to significantly improve the reliability and timeliness of temperature-related fault warnings. It can provide early warnings based on the characteristics of the event sequence when an abnormality is about to occur, and provides an innovative technical path based on positive monitoring for the preventive maintenance of the temperature of the core components of wind turbines, which is a significant improvement.

[0039] It should be noted that the present application replaces the statistical model with the coupling characteristics obtained through physical mechanism rules and calculations, which can directly reflect the physical coupling relationship between the equipment operating status and the cooling system, thereby improving the accuracy of the variable pitch drive overheating and cooling pump failure scenarios. Before the temperature reaches the absolute threshold, the coupling characteristics can be used to predict the decrease in heat dissipation efficiency in advance, thereby avoiding fault escalation.

[0040] It should be noted that by detecting physically infeasible events and removing data that violates basic laws, we can extract physically credible, logically self-consistent, and temporally coherent information flows, thus providing high-quality input for subsequent analysis and improving the reliability of subsequent analysis.

[0041] It should be noted that by dynamically allocating weights to associate abnormality types with physical components, the root cause can be accurately located, which can reduce the troubleshooting costs for operation and maintenance personnel.

[0042] It should be noted that by defining the coupling characteristics, the physical essence of the relationship between the device heat dissipation efficiency and the load temperature rise can be directly mapped, the states of the mechanical system and the electrical system can be traced, and the problem that the statistical characteristics of the traditional method have no actual physical meaning is solved; the dynamic window mechanism improves the transient anomaly sensitivity and can avoid the response delay of the fixed window to emergencies.

[0043] It should be noted that the physical mechanism rules include: In the generator temperature control module, the generator temperature is positively correlated with the generator speed. When the generator is at a high temperature, the lubricating oil becomes thinner, the friction force decreases, and the generator speed increases; when the generator temperature becomes lower, the lubricating oil becomes thicker, the friction force increases, and the generator speed decreases. In the cooling system module, when the generator temperature rises, the water temperature at the water-cooled outlet valve rises and the pump and fan are in an active state. In the electrical load module, when the active power of the converter or the generator speed is too high, the temperature of the key components strongly related to the cooling system efficiency can be controlled. In the mechanical state module, the bearing temperature is inversely proportional to the bearing vibration amplitude, and the heat dissipation efficiency decreases when the hydraulic system pressure is abnormal.

[0044] In a specific embodiment, for the abnormal temperature of the front bearing of the generator, the system makes the following judgments: In the time window 1 (08:00:00 - 08:00:30), the Scada data is shown in Table 1. The temperature of the front bearing of the key device, the generator, rises from the initial 30°C to 34°C, the temperature change is 4°C, the average value is 32°C, the temperature of the cooling medium (average water temperature at the water-cooled outlet valve) is 28°C, the minimum value is taken as 0.01, and the change in the active power of the converter is 100KW. Calculate the coupling characteristics of the Scada data. The device-cooling temperature difference ratio TCR = 32 / (28 + 0.01) = 1.14. According to the linear regression slope formula The calculated linear regression slope is 0.14 °C / second, without an obvious sudden increase, which conforms to the steady-state temperature rise. The load-temperature gradient matching degree LTM = 4 / 100 = 0.04; The average temperature of the front bearing of the generator is 32°C, which is lower than the preset bearing temperature threshold of 45°C. The temperature difference between the inner ring and the outer ring of the front bearing of the generator is 15°C, which is less than the preset bearing temperature difference threshold of 20°C. The physical constraint conflict flag PCC = 0; The average water temperature at the water-cooled outlet valve is 28°C, which is lower than the preset water-cooled temperature threshold of 40°C. The temperature difference between the water temperature at the water-cooled outlet valve and the water temperature at the water-cooled inlet valve is 6°C, which is greater than the preset water-cooled temperature difference threshold of 5°C. The physical constraint conflict flag PCC = 0; The ambient average temperature is 25°C, the temperature threshold of the high-temperature weather warning line is 35°C, the cabin temperature is 35°C, the forced ventilation failure threshold is 50°C. If the ambient temperature and the cabin temperature are lower than the preset threshold, then the physical constraint conflict flag PCC = 0; The current window does not trigger the physical mechanism rule. The weight assigned to the coupling feature is 0.1, and the anomaly score for window period 1 is 0.1×(1.14 + 0.04 + 0) = 0.118; Table 1: Scada data for window period 1

[0045] The generator Scada data for window period 2 (08:00:10 - 08:00:40) is shown in Table 2. The temperature of the key equipment, the generator, rises from the initial 34°C to 61°C, with a temperature change of 27°C. Taking the average value of 47.5°C, the average temperature of the corresponding cooling medium (average water-cooled outlet valve water temperature) is 41°C, with the minimum value taken as 0.01, and the power change is 100 KW; Calculating the coupling feature of the Scada data, the equipment-cooling temperature difference ratio TCR = 47.5 / (41 + 0.01) = 1.16, which has increased by 1.02 times compared to window period 1 and is less than twice the preset increase threshold, so the physical mechanism rule is not triggered; The load-temperature gradient matching degree LTM = 27 / 100 ≈ 0.27, which is less than the preset warning threshold of 0.5, so the physical mechanism rule is not triggered; The average temperature of the generator front bearing is 47.5°C, which is higher than the preset front bearing temperature threshold of 45°C. The temperature difference between the inner and outer rings of the generator front bearing is 25°C, which is greater than the preset bearing temperature difference threshold of 20°C, so the physical constraint conflict flag PCC = 1; The average water-cooled outlet valve water temperature is 41°C, which is higher than the preset water-cooling temperature threshold of 40°C. The temperature difference between the water-cooled outlet valve water temperature and the water-cooled inlet valve water temperature is 13°C, which is greater than the preset water-cooling temperature difference threshold of 5°C, so the physical constraint conflict flag PCC = 1; The ambient average temperature is 25°C, the temperature threshold of the high-temperature weather warning line is 35°C, the cabin temperature is 35°C, the forced ventilation failure threshold is 50°C. If the ambient temperature and the cabin temperature are lower than the preset threshold, then the physical constraint conflict flag PCC = 0; The current window only triggers 1 physical mechanism rule. The weight assigned to the coupling feature is 0.3, and the anomaly score is 0.3×(1.16 + 0.27 + 2) = 1.029; Table 2: Scada data for window period 2

[0046] During window period 3 (08:00:20 - 08:00:50), the Scada data is shown in Table 3. The temperature of the key equipment generator rises from the initial 42°C to 72°C, with a temperature change of 30°C. Taking the average value of 57°C, the average temperature of the corresponding cooling medium (water-cooled outlet valve water temperature) is 45°C, the minimum value is taken as 0.01, and the power change is 30 KW; Calculate the coupling characteristics of the Scada data. The equipment-cooling temperature difference ratio TCR = 57 / (45 + 0.01) = 1.27, which has increased by 1.09 times compared to window period 2. However, with an increase in three consecutive window periods, the heat dissipation efficiency decays, triggering the physical mechanism rule; The load-temperature gradient matching degree LTM = 30 / 30 = 1 > 0.5, triggering the physical mechanism rule; The temperature of the front bearing of the generator is 57°C, which is higher than the preset front bearing temperature threshold of 45°C. The temperature difference between the inner ring and the outer ring of the front bearing of the generator is 25°C, which is greater than the preset bearing temperature difference threshold of 20°C. The physical constraint conflict flag PCC = 1; The average water-cooled outlet valve water temperature is 45°C, which is higher than the preset water-cooled temperature threshold of 40°C. The temperature difference between the water-cooled outlet valve water temperature and the water-cooled inlet valve water temperature is 7°C, which is greater than the preset water-cooled temperature difference threshold of 5°C. The physical constraint conflict flag PCC = 1; The average ambient temperature is 36°C, the temperature threshold of the high-temperature weather warning line is 35°C, the cabin temperature is 55°C, and the forced ventilation failure threshold is 50°C. Since the ambient temperature and the cabin temperature are higher than the preset thresholds, the physical constraint conflict flag PCC = 1; In the current window, 3 physical mechanism rules of the core equipment are triggered. The weight assigned to the coupling characteristics is 0.6, and the anomaly score is 0.6×(1.27 + 1 + 3) = 3.162; Table 3: Scada data for window period 3

[0047] Assume that the preset reference threshold is 4. The anomaly score within three window periods is 4.309, which is greater than the preset reference threshold and the number threshold of 3 for the preset physical constraint conflict flag PCC being 1, triggering a red alert.

[0048] The following refers to Figure 2 , Figure 2 which shows a fan high-temperature anomaly detection system based on time-series feature fusion, including a data processing module a, a mechanism modeling module b, a coupling module c, and a warning module d.

[0049] In a specific embodiment, the data processing module a is configured to: obtain the Scada data of the wind turbine set for downsampling processing, process the boundary instantaneous values in the Scada data through a trend quantization method, and then perform smoothing high-frequency noise processing through moving average filtering to extract the temperature characteristics and operating state characteristics of the Scada data.

[0050] In a specific embodiment, the mechanism modeling module b is configured to: divide the Scada data into a generator temperature control module, a cooling system module, an electrical load module, and a mechanical state module, and associate the temperature characteristics and operating state characteristics of the Scada data with physical mechanism rules for mechanism modeling.

[0051] In a specific embodiment, the coupling module c is configured to: combine mechanism modeling or calculation to obtain the coupling characteristics of the Scada data, where the coupling characteristics include the equipment-cooling temperature difference ratio TCR, the load-temperature gradient matching degree LTM, and the physical constraint conflict flag PCC.

[0052] In a specific embodiment, the warning module d is configured to: assign different weights to the coupling characteristics based on the coupling strength for anomaly scoring, and perform multi-level warnings based on the anomaly scoring. When the anomaly scoring is greater than a preset basic threshold, a yellow warning is issued. When the anomaly scoring is greater than the preset basic threshold and the number of physical constraint conflict flags PCC with a value of 1 exceeds a preset number threshold, a red warning is issued; The anomaly scoring formula is: , where TCR represents the equipment-cooling temperature difference ratio, LTM represents the load-temperature gradient matching degree, PCC represents the number of physical constraint conflict flags with a value of 1, Score represents the anomaly scoring, and Weight represents the different weights assigned to the coupling characteristics.

[0053] It should be noted that this application performs high-precision anomaly detection through physical mechanism rules, directly reflects the physical coupling relationship between the equipment operating state and the cooling system through coupling characteristics, directly associates the anomaly type and physical components through dynamic weight assignment, can avoid the performance degradation of the wind turbine caused by historical dependence, achieve accurate root cause positioning, reduce the operation and maintenance troubleshooting cost, filter false alarms caused by sensor noise or short-term load fluctuations, reduce the false alarm rate, and improve the accuracy rate.

[0054] It should be noted that the window length and threshold benchmark of this application are adjusted according to the real-time working conditions, which can avoid performance degradation caused by historical dependence, reverse position the specific fault location through weight combination, and directly map the anomaly to the operation and maintenance actions.

[0055] It should be noted that the real-time health assessment method of industrial equipment based on thermodynamic characteristics proposed in this application innovatively presents a two-dimensional coupling analysis model of the equipment-cooling temperature difference ratio TCR and the load-temperature gradient matching degree. The equipment-cooling temperature difference ratio TCR continuously quantifies the attenuation characteristics of heat dissipation efficiency by comparing the temperature gradient distribution of key nodes in the heat exchange system; the load-temperature gradient matching degree LTM constructs a time-domain criterion for the non-steady temperature rise trend based on the dynamic process detection of temperature change kinetics. The two parameters act synergistically on the research of the thermal load evolution law of the core components of the fan rotating machinery. By observing the time-frequency dimension extension of the continuous window period, the early signs of equipment thermal anomalies are captured without the intervention of prediction algorithms. Compared with the traditional thermal failure warning technology based on prediction models, this application significantly improves the reliability and timeliness of temperature-related fault warnings by means of the intrinsic analysis of the time series of SCADA data, and can give early warnings according to the characteristics of the event sequence when an anomaly is about to occur, providing an innovative technical path for the preventive maintenance of the temperature of the core components of the wind turbine based on forward monitoring, which is one of the innovation points of this application.

[0056] Next, refer to Figure 3 , which shows a schematic structural diagram of a computer system of an electronic device suitable for implementing the embodiments of the present application. Figure 3 The shown electronic device is only an example and should not impose any restrictions on the functions and usage scope of the embodiments of the present application.

[0057] As Figure 3 shown, the computer system includes a central processing unit (CPU) 301, which can perform various appropriate actions and processes according to the program stored in the read-only memory (ROM) 302 or the program loaded from the storage section 308 into the random access memory (RAM) 303. In the RAM 303, various programs and data required for system operation are also stored. The CPU 301, ROM 302, and RAM 303 are connected to each other via a bus 304. The input / output (I / O) interface 305 is also connected to the bus 304.

[0058] The following components are connected to the I / O interface 305: an input section 306 including a keyboard, a mouse, etc.; an output section 307 including, for example, a liquid crystal display (LCD), etc., and a speaker, etc.; a storage section 308 including a hard disk, etc.; and a communication section 309 including a network interface card such as a LAN card, a modem, etc. The communication section 309 performs communication processing via a network such as the Internet. A drive 310 is also connected to the I / O interface 305 as needed. A removable medium 311, such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, etc., is installed on the drive 310 as needed so that the computer program read from it can be installed into the storage section 308 as needed.

[0059] In particular, according to an embodiment of the present disclosure, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, an embodiment of the present disclosure includes a computer program product that includes a computer program carried on a computer-readable storage medium, and the computer program includes program code for performing the methods shown in the flowcharts. In such an embodiment, the computer program can be downloaded and installed from the network through the communication section 309, and / or installed from the removable medium 311. When the computer program is executed by the central processing unit (CPU) 301, the above-described functions defined in the method of the present application are performed. It should be noted that the computer-readable storage medium of the present application can be a computer-readable signal medium, a computer-readable storage medium, or any combination of the two. The computer-readable storage medium can be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination of the above. More specific examples of the computer-readable storage medium can include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In the present application, the computer-readable storage medium can be any tangible medium that contains or stores a program, and the program can be used by or in combination with an instruction execution system, apparatus, or device. In the present application, the computer-readable signal medium can include a data signal propagated in a baseband or as part of a carrier wave, in which the computer-readable program code is carried. Such a propagated data signal can take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. The computer-readable signal medium can also be any computer-readable storage medium other than the computer-readable storage medium, and the computer-readable storage medium can send, propagate, or transmit a program for use by or in combination with an instruction execution system, apparatus, or device. The program code contained on the computer-readable storage medium can be transmitted by any appropriate medium, including but not limited to: wireless, wire, optical fiber, RF, etc., or any suitable combination of the above.

[0060] Computer program code for performing the operations of this application can be written in one or more programming languages or combinations thereof. The programming languages include object-oriented programming languages such as Java, Smalltalk, C++, and also include conventional procedural programming languages such as the "C" language or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, executed as a stand-alone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In the case of a remote computer, the remote computer can be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or can be connected to an external computer (e.g., by using an Internet service provider to connect through the Internet).

[0061] The flowcharts and block diagrams in the accompanying drawings illustrate the possible architectures, functions, and operations of systems, methods, and computer program products according to various embodiments of the present application. In this regard, each block in the flowchart or block diagram may represent a module, a program segment, or a part of code that contains one or more executable instructions for implementing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the blocks may occur in a different order than marked in the accompanying drawings. For example, two consecutive blocks shown may actually be executed substantially in parallel, and they may sometimes be executed in the reverse order, depending on the functions involved. It should also be noted that each block in the block diagram and / or flowchart, and combinations of blocks in the block diagram and / or flowchart, can be implemented by a dedicated hardware-based system that performs the specified functions or operations, or can be implemented by a combination of dedicated hardware and computer instructions.

[0062] The modules described in the embodiments of the present application can be implemented in software or in hardware.

[0063] As another aspect, the present application also provides a computer-readable storage medium, which may be included in the electronic device described in the above embodiments; or may exist alone without being assembled into the electronic device. The above computer-readable storage medium carries one or more programs. When the above one or more programs are executed by the electronic device, the electronic device is enabled to: obtain the Scada data of the wind turbine set for downsampling processing, process the boundary instantaneous values in the Scada data through a trend quantization method, and then perform smoothing high-frequency noise processing through a moving average filter to extract the temperature characteristics and operating state characteristics of the Scada data; divide the Scada data into a generator temperature control module, a cooling system module, an electrical load module, and a mechanical state module, and associate the temperature characteristics and operating state characteristics of the Scada data with physical mechanism rules to perform mechanism modeling and combine mechanism modeling or calculation to obtain the coupling characteristics of the Scada data, where the coupling characteristics include the equipment-cooling temperature difference ratio TCR, the load-temperature gradient matching degree LTM, and the physical constraint conflict flag PCC; assign different weights to the coupling characteristics based on the coupling strength for anomaly scoring, and perform multi-level early warning based on the anomaly scoring. When the anomaly scoring is greater than a preset basic threshold, a yellow early warning is issued. When the anomaly scoring is greater than the preset basic threshold and the number of 1s in the physical constraint conflict flag PCC exceeds a preset number threshold, a red early warning is issued.

[0064] Finally, it should be noted that the above description is only a preferred embodiment of the present application and an explanation of the applied technical principles. Those skilled in the art should understand that the scope of the invention involved in the present application is not limited to the technical solutions formed by the specific combination of the above technical features, but should also cover other technical solutions formed by any combination of the above technical features or their equivalent features without departing from the above inventive concept. For example, the technical solutions formed by mutually replacing the above features with the (but not limited to) technical features with similar functions disclosed in the present application.

Claims

1. A method for detecting abnormal high temperature of a fan based on temporal feature fusion, characterized in that, Including: S1: Obtain the Scada data of the wind turbine set for downsampling processing. Process the boundary instantaneous values in the Scada data through a trend quantization method, and then perform smoothing high-frequency noise processing through moving average filtering to extract the temperature characteristics and operating state characteristics of the Scada data; S2: Divide the Scada data into a generator temperature control module, a cooling system module, an electrical load module, and a mechanical state module, and associate the temperature characteristics and operating state characteristics of the Scada data with physical mechanism rules for mechanism modeling; S3: Combine mechanism modeling or calculation to obtain the coupling characteristics of the Scada data. The coupling characteristics include the equipment-cooling temperature difference ratio TCR, the load-temperature gradient matching degree LTM, and the physical constraint conflict flag PCC; S4: Assign different weights to the coupling characteristics based on the coupling strength for anomaly scoring, and perform multi-level early warning based on the anomaly scoring. When the anomaly scoring is greater than the preset basic threshold, a yellow early warning is issued. When the anomaly scoring is greater than the preset basic threshold and the number of the physical constraint conflict flag PCC being 1 exceeds the preset number threshold, a red early warning is issued; The anomaly scoring formula is: ， Wherein, TCR represents the equipment-cooling temperature difference ratio, LTM represents the load-temperature gradient matching degree, PCC represents the number of the physical constraint conflict flag being 1, Score represents the anomaly scoring, and Weight represents the different weights assigned to the coupling characteristics.

2. The method according to claim 1, wherein The temperature characteristics include the pitch drive temperature, the ambient temperature, the water temperature at the water-cooling outlet valve, the water temperature at the water-cooling inlet valve, the nacelle temperature, the generator bearing temperature, and the hub temperature; The operating state characteristics include the pump state and the fan state.

3. The method according to claim 1, wherein Step S1 further includes filtering the Scada data with the temperature characteristics exceeding the physical limit threshold, cleaning the Scada data with logical conflicts, replacing the missing values with the mean value of the Scada data, and performing sliding window time series processing on the Scada data.

4. The method according to claim 1, characterized in that, The calculation formula of the equipment-cooling temperature difference ratio TCR is: ， Among them, TCR represents the equipment-cooling temperature difference ratio, represents the temperature of the key equipment, represents the temperature of the corresponding cooling medium, represents the minimum value.

5. The method according to claim 1, characterized in that, The calculation formula of the load-temperature gradient matching degree LTM is: ， Among them, LTM represents the load-temperature gradient matching degree, represents the temperature change of the device within the window, represents the change in load power, represents the minimum value.

6. The method according to claim 1, wherein Define the physical constraint conflict flag PCC based on the system physical mechanism rules, including: If the temperature of the front bearing of the generator is higher than the preset front bearing temperature threshold, and the temperature difference between the inner ring temperature and the outer ring temperature of the front bearing or the rear bearing of the generator exceeds the preset bearing temperature difference threshold, then the physical constraint conflict flag PCC is 1; If the water temperature at the water-cooling outlet valve is higher than the preset water-cooling temperature threshold, and the temperature difference between the water temperature at the water-cooling outlet valve and the water temperature at the water-cooling inlet valve exceeds the preset water-cooling temperature difference threshold, then the cooling system efficiency decreases, and the physical constraint conflict flag PCC is 1; If the ambient temperature is higher than the temperature threshold of the high-temperature weather warning line, and the nacelle temperature is higher than the preset forced ventilation failure threshold, then there is an abnormality in the ambient or nacelle temperature, and the physical constraint conflict flag PCC is 1; Otherwise, the physical constraint conflict flag PCC is 0.

7. The method according to claim 1, wherein The division of the Scada data into a generator temperature control module, a cooling system module, an electrical load module, and a mechanical state module is specifically as follows: The generator temperature control module includes the temperature of the front bearing of the generator, the temperature of the rear bearing of the generator, the temperature of the inner ring of the front bearing of the generator, and the temperature of the outer ring of the front bearing of the generator; The cooling system module includes the water temperature of the water-cooled outlet valve, the water temperature of the water-cooled inlet valve, the pump status, and the fan status; The electrical load module includes the active power of the converter and the generator speed; The mechanical state module includes the pressure of the hydraulic system and the bearing vibration amplitude.

8. The method according to claim 1, wherein Assigning different weights to the coupling characteristics based on the coupling strength includes: If two or more physical mechanism rules are triggered in the current window, a high weight is assigned to the coupling characteristics; If one physical mechanism rule is triggered in the current window, a medium weight is assigned to the coupling characteristics; If no physical mechanism rule is triggered in the current window, a low weight is assigned to the coupling characteristics.

9. A high-temperature anomaly detection system for a fan based on temporal feature fusion, characterized in that, Including: Data processing module: Obtain the Scada data of the wind turbine set for downsampling processing, process the boundary instantaneous values in the Scada data through a trend quantization method, and then perform smoothing of high-frequency noise through a moving average filter to extract the temperature characteristics and operation state characteristics of the Scada data; Mechanism modeling module: Divide the Scada data into a generator temperature control module, a cooling system module, an electrical load module, and a mechanical state module, and associate the temperature characteristics and operation state characteristics of the Scada data with physical mechanism rules for mechanism modeling; Coupling module: Combine mechanism modeling or calculation to obtain the coupling characteristics of the Scada data, and the coupling characteristics include the equipment-cooling temperature difference ratio TCR, the load-temperature gradient matching degree LTM, and the physical constraint conflict flag PCC; Early warning module: Assign different weights to the coupling characteristics based on the coupling strength for anomaly scoring, and perform multi-level early warning based on the anomaly scoring. When the anomaly scoring is greater than a preset basic threshold, a yellow early warning is issued. When the anomaly scoring is greater than the preset basic threshold and the number of physical constraint conflict flags PCC being 1 exceeds a preset number threshold, a red early warning is issued; The anomaly scoring formula is: ， Among them, TCR represents the equipment-cooling temperature difference ratio, LTM represents the load-temperature gradient matching degree, PCC represents the number of physical constraint conflict flags being 1, Score represents the anomaly scoring, and Weight represents the different weights assigned to the coupling characteristics.

10. A computer program product having one or more computer programs thereon, characterized in that, When the computer program is executed by a computer processor, the method described in any one of claims 1-8 is implemented.

Images